Article
Astronomy & Astrophysics
Karl Daningburg, Richard O'Shaughnessy
Summary: Gravitational wave science relies on expensive numerical simulations, which limits exploration of the binary black hole parameter space. We have developed a new acquisition function that reduces the cost of simulations.
Article
Chemistry, Physical
Takafumi Shiraogawa, Jun-ya Hasegawa
Summary: In this study, a method is proposed for designing functional molecules in chemical space using computational quantum alchemy, which explicitly considers the geometric stability of molecules. The proposed method allows simultaneous prediction of functional molecules and their target desired properties in an inverse design fashion.
JOURNAL OF PHYSICAL CHEMISTRY LETTERS
(2022)
Article
Mathematics, Applied
Mihaela Pricop-Jeckstadt
Summary: This paper investigates the minimax rates of convergence for a class of linear inverse problems with correlated noise, discussing the phase transition phenomenon in functional data analysis and computing thresholds for different smoothness conditions. The high cost of data correlation is highlighted.
Article
Computer Science, Artificial Intelligence
Li Song, Zhou Ge, Edmund Y. Lam
Summary: This paper proposes a novel inverse imaging scheme, called D-ADMM, by combining duality and the alternating direction method of multipliers (ADMM). By solving the dual problem, this method can find the estimated nontrivial lower bound of the objective function and guide the primal iterations. In image super-resolution, D-ADMM-TV shows comparable or slightly better results compared to other methods.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Mathematics, Applied
Can Tong, Yueyang Teng, Yudong Yao, Shouliang Qi, Chen Li, Tie Zhang
Summary: In this paper, an eigenvalue-free iterative shrinkage threshold algorithm (EFISTA) based on majorization-minimization is proposed to avoid the computation of eigenvalues, showing better performance in large-scale problems. The algorithm can be extended to a fast EFISTA and proofs of convergence and convergence rate are provided, with experimental results demonstrating its effectiveness and feasibility.
Article
Mathematics, Applied
Nguyen Trung Thanh
Summary: This paper deals with a 1D time-domain inverse scattering problem for the Helmholtz equation, which aims to determine penetrable scatterers based on boundary measurements of the scattering data. It is formulated as a coefficient identification problem for a wave equation. By utilizing the Laplace transform, the inverse problem is converted into an overdetermined nonlinear system of partial differential equations. A Carleman weighted objective functional, proven to be strictly convex in an arbitrary set in a Hilbert space, is constructed to solve this system. An alternating minimization algorithm is employed to minimize the Carleman weighted objective functional. Numerical results are presented to demonstrate the effectiveness of the proposed algorithm.
Article
Chemistry, Physical
Haoran He, Randall J. Meyer, Robert M. Rioux, Michael J. Janik
Summary: Cyclohexene is a chemical intermediate produced through catalytic partial hydrogenation of benzene. Density functional theory calculations and microkinetic modeling illustrate that the binding energy of benzene is a predictor of a catalyst's cyclohexene selectivity. Ni5Ga3 and Ni3Ga intermetallic compounds are predicted to be highly selective catalysts for benzene hydrogenation to cyclohexene.
Article
Engineering, Multidisciplinary
Yuehan Yang, Hu Yang
Summary: This study proposes a new adaptive and reversed penalty method for dealing with correlated data to achieve stable and accurate estimation; the effectiveness and success of the method are verified through simulations and application to financial data.
APPLIED MATHEMATICAL MODELLING
(2021)
Article
Geochemistry & Geophysics
Lingqian Wang, Hui Zhou, Hanming Chen, Yufeng Wang, Yuanpeng Zhang
Summary: In this article, a structurally constrained multichannel L1-2 minimization method is proposed for seismic attenuation compensation. This method shows better stability and accuracy compared to traditional compensation methods and enhances the spatial continuity of the compensated results.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Mathematics, Applied
A. Buccini, P. Diaz de Alba
Summary: This paper presents a regularization method for reconstructing the soil structure using non-invasive techniques, such as frequency domain electromagnetic induction devices. The method solves a variational problem to determine the electrical conductivity of the ground and balances the data fitting term and regularization term to obtain an accurate solution. Results from synthetic and real data demonstrate the good performance of the proposed method.
Article
Nuclear Science & Technology
M-h. Aumeunier, M. Le Bohec, R. Brunet, A. Juven, M. Adel, X. Artusi, R. Miorelli, C. Reboud, F. Rigollet
Summary: Infrared thermography is an essential diagnostic tool in fusion devices, but qualitative and quantitative analysis are challenging due to disturbance phenomena in metallic environments. This paper develops inverse methods to retrieve real surface temperature by considering variable emissivity and filtering reflections, using gradient methods and machine learning techniques. The accuracy of temperature estimation is improved by more than 6% compared to usual methods.
NUCLEAR MATERIALS AND ENERGY
(2022)
Article
Computer Science, Artificial Intelligence
Iciar Llorens Jover, Thomas Debarre, Shayan Aziznejad, Michael Unser
Summary: This article describes the inverse problem of constructing sparse parametric continuous curve models that fit a sequence of contour points, and introduces the prior knowledge and solution method involved. The performance of the model in contour reconstruction is verified through experiments.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Mechanical
M. Aucejo, O. De Smet
Summary: Input estimation remains a significant issue in structural dynamics, with two main groups of inverse methods in time and frequency domains. This paper introduces a generalized multiplicative regularization for estimating mechanical loads on linear structures, demonstrating high solution accuracy through numerical and real-world applications. The extra tuning parameter in this approach plays a key role in enhancing results amidst measurement noise levels.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Mathematics, Applied
Tina Mai, Daniele Mortari
Summary: This paper introduces an efficient approach to solve quadratic and nonlinear programming problems subject to linear equality constraints using the theory of functional connections. The approach does not rely on the traditional Lagrange multiplier technique, but instead utilizes the free vector g and combines Newton's method with an elimination scheme for solving the problems.
JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS
(2022)
Article
Green & Sustainable Science & Technology
Przemyslaw Kowalik, Magdalena Rzemieniak
Summary: The importance of pump scheduling in improving energy efficiency, production costs, emissions, and reliability is discussed. The objective is to minimize the cost of electric power used by water-supplying pumps through optimization and flexible management of the pump system.
Article
Mathematics, Applied
Wenjie Shi, Daniel M. Tartakovsky
Summary: This article investigates the impact of stiffness of random ODEs on the performance of gPC and introduces the use of gPC with parallel MIRK schemes for solving stiff random ODEs. The stiffness analysis provides a direct way to determine whether a random ODE and its corresponding gPC equations are stiff. This theoretical analysis plays a key role in designing numerical implementations and other stochastic computation methods.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2022)
Article
Green & Sustainable Science & Technology
Bulbul Ahmmed, Velimir V. Vesselinov
Summary: This study identifies multiple geothermal prospects in the Great Basin region based on shallow groundwater chemical data. Using machine learning analysis, the study finds hidden geothermal features and probabilities, which are consistent with traditional geothermal potential analysis. The study also provides continuous attribute predictions for future detailed geothermal exploration.
Article
Physics, Multidisciplinary
Adam Rupe, Velimir V. Vesselinov, James P. Crutchfield
Summary: This article explains the success of data-driven models in partially-observed systems. Data-driven models use techniques such as delay-coordinate embeddings to implicitly model the effects of missing degrees of freedom, outperforming physics-based models in systems with few observable degrees of freedom.
NEW JOURNAL OF PHYSICS
(2022)
Article
Physics, Applied
Abhishek Chandra, Bram Daniels, Mitrofan Curti, Koen Tiels, Elena A. Lomonova, Daniel M. Tartakovsky
Summary: This article presents an approach for modelling hysteresis in piezoelectric materials using sparse regression techniques. The study demonstrates the efficiency and accuracy of the proposed approach through numerical experiments and comparisons with traditional regression-based and neural network methods. The source code is available for further exploration and implementation.
APPLIED PHYSICS LETTERS
(2023)
Article
Energy & Fuels
Maruti K. Mudunuru, Bulbul Ahmmed, Elisabeth Rau, Velimir V. Vesselinov, Satish Karra
Summary: Geothermal energy is recognized as an important renewable resource for generating flexible electricity. The Tularosa Basin in New Mexico has been identified as having significant geothermal potential through play fairway analysis funded by the US Department of Energy. A machine learning methodology called non-negative matrix factorization with custom k-means clustering has been used to identify potential geothermal sites in the basin, utilizing the open-source ML framework GeoThermalCloud (GTC). This ML analysis aligns with existing studies and enhances confidence in the GTC framework for accelerating geothermal exploration and resource development.
Article
Electrochemistry
Weiyu Li, Hamdi A. Tchelepi, Daniel M. Tartakovsky
Summary: The study analyzed the role of buffer layer materials in lithium batteries and identified the conditions under which the buffer layer stabilizes electrodeposition and suppresses dendrite growth. The model predicted the effectiveness of several prospective buffer materials in stabilizing electrodeposition and suppressing dendrite growth, which was consistent with experimental findings. This has important implications for guiding the experimental and computational discovery of new buffer materials.
JOURNAL OF THE ELECTROCHEMICAL SOCIETY
(2023)
Article
Environmental Sciences
Weiyu Li, Daniel M. Tartakovsky
Summary: Accurately estimating evapotranspiration is crucial for smart agriculture and sustainable groundwater management. We present novel methods, EnKF and MLE, that can infer spatially varying ET rates and root water uptake profiles from soil-moisture measurements. Our methods accurately estimate total ET rates and root-uptake profiles in a drip irrigation setting and are up to two orders of magnitude faster than the standard EnKF.
WATER RESOURCES RESEARCH
(2023)
Article
Mathematics, Applied
Hannah Lu, Francesco Giannino, Daniel M. Tartakovsky
Summary: Mathematical models play a key role in estimating patient-specific initial viral load, predicting the course of infection, etc. The development of COVID-19 pandemics has led to increasingly complex models. We found that the widely used Target Cell Limited model fails the identifiability test, but we propose an identifiable and parsimonious model that matches observations and predictions of more complex counterparts.
APPLIED MATHEMATICS LETTERS
(2023)
Article
Computer Science, Interdisciplinary Applications
Hannah Lu, Daniel M. Tartakovsky
Summary: DRIPS is a new model reduction framework that combines offline local model reduction with online parameter interpolation. By using dynamic mode decomposition to build a low-rank linear surrogate model, it is able to directly model quantities of interest and has higher computational efficiency compared to traditional proper orthogonal decomposition methods.
JOURNAL OF COMPUTATIONAL PHYSICS
(2023)
Article
Electrochemistry
Weiyu Li, Daniel M. Tartakovsky
Summary: This study presents a homogenized thermal model for a spherical active particle coated with a carbon binder domain (CBD) immersed in a liquid electrolyte. The model replaces the composite particle with a homogeneous particle with equivalent thermal conductivity, preserving the amount of released heat at the solid/electrolyte interface. This analytical expression for thermal conductivity can be readily integrated into thermal simulations, providing a means to account for CBD in battery design and management models.
JOURNAL OF THE ELECTROCHEMICAL SOCIETY
(2023)
Article
Energy & Fuels
Livia Paiva Fulchignoni, Daniel M. Tartakovsky
Summary: Equations of state (EoS) are crucial for modeling fluid mixtures' phase equilibrium, but their parameterization involves fitting models to experimental data via nonlinear, non-convex, multivariate optimization. We show that subjective choices of optimization algorithms and initial guesses impact EoS predictions, resulting in fundamental uncertainties even after tuning to limited experimental data. Using two hydrocarbon reservoir fluids as examples, we demonstrate dramatic differences in predicting the fluids' thermophysical behavior in unsampled pressure and temperature regions depending on EoS parameterizations. We propose a probabilistic treatment of design variables to quantify the predictive uncertainty of resulting fluid models.
GEOENERGY SCIENCE AND ENGINEERING
(2023)
Article
Energy & Fuels
Livia Paiva Fulchignoni, Christiano Garcia da Silva Santim, Daniel M. Tartakovsky
Summary: Offshore development requires significant investments that are subject to multiple uncertainties. Quantifying the uncertainties in reservoir production predictions and project revenue can mitigate risks and enable more informed business decisions.
GEOENERGY SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Interdisciplinary Applications
Apoorv Srivastava, Wei Kang, Daniel M. Tartakovsky
Summary: This paper introduces a mathematical formulation of feature-informed data assimilation (FIDA), which utilizes the information about feature events in dynamical systems to estimate state variables and unknown parameters. The observation operator in FIDA is a set-valued functional, which is different from conventional data assimilation. Through three numerical experiments, FIDA's ability to estimate model parameters from noisy observations is demonstrated.
JOURNAL OF COMPUTATIONAL PHYSICS
(2023)
Article
Computer Science, Information Systems
Daniel Dylewsky, David Barajas-Solano, Tong Ma, Alexandre M. Tartakovsky, J. Nathan Kutz
Summary: This article introduces a novel load forecasting method using Dynamic Mode Decomposition and Gaussian Process Regression to model the dynamics of grid load. It is compared to other forecasting techniques and shows superior performance. The method offers advantages in terms of interpretability and parsimony, and is important for load forecasting in power systems engineering.
Article
Mathematics, Applied
Wenjie Shi, Daniel M. Tartakovsky
Summary: The study investigates the impact of stiffness of random ODEs on gPC performance and introduces gPC with parallel MIRK schemes to solve random stiff ODEs. The stiffness analysis and computational experiments validate the feasibility and effectiveness of the method in solving random ODEs.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2022)